The method of assessing shape similarity of object-based classification result of remote sensing imagery

Abstract

This paper is about the shape-based geometric accuracy index, which is an important evaluation index of the object-based classification result. There are three important features used in developing geometric accuracy, area, position and shape. The feature of shape can reflect the spatial pattern of classified object, which is an important target of object-based classification. For this reason, we use shape similarity to evaluate the geometric accuracy of object-based classification result of remote sensing imagery. The evaluation method is developed from slice model. The effectiveness of this method has been proved in case study.